A mobile robot in an urban environment has to deal with many potential hazards. In this paper, we propose a fast stereo vision based mapping algorithm for detecting ramps, drop-offs and obstacles. Our algorithm builds a hybrid 3D model of the robot's local surroundings that consists of a 3D occupancy grid, used for modeling obstacles and unsafe regions, and of planes, for modeling the ground and potentially traversable regions. The hybrid 3D model is analyzed to identify safe regions and the safety information is captured in an annotated 2D grid map called a local safety map that can be used by the robot to plan safe local paths. We evaluate our algorithm comprehensively by testing it in varied environments and comparing the results to laser range-finder based ground truth data. We believe the evaluation framework introduced here is also useful for evaluating the performance of other mapping algorithms and towards that purpose we plan to make our datasets and related ground truth data publicly available.

PDF, HTML

In International Conference on Intelligent Robots and Systems (IROS), 2009.

@InProceedings{Murarka-iros-09, title={A stereo vision based 3D mapping algorithm for detecting ramps, drop-offs, and obstacles for safe local navigation}, author={Aniket Murarka and Benjamin Kuipers}, booktitle={International Conference on Intelligent Robots and Systems (IROS)}, url="http://www.cs.utexas.edu/users/ai-lab/pub-view.php?PubID=126850", year={2009} }